. 2014 Feb 25;10(4):1563–1575. doi: 10.1021/ct401111c

Table 5. Linear Scaling Factors for SAPT0/aug-cc-pVTZ Energy Terms^a.

	fSAPT0(1)	fSAPT0(2)	fSAPT0(3)
E_elst,r⁽¹⁰⁾	1.01 ± 0.02	1.00^b	1.00^b
E_exch⁽¹⁰⁾	1.02 ± 0.02	1.00^b	1.00^b
E_ind,r⁽²⁰⁾	0.76 ± 0.08	1.00^b	1.00^b
E_exch–ind,r⁽²⁰⁾	0.70 ± 0.08	1.00^b	1.00^b
δE_HF,r⁽²⁾	1.06 ± 0.08	1.00^b	1.00^b
E_disp⁽²⁰⁾	0.93 ± 0.01	0.96 ± 0.02	0.76^c ± 0.01
E_exch–disp⁽²⁰⁾	1.7 ± 0.2	2.1 ± 0.2	0.76^c ± 0.01
test RMSE	0.66 ± 0.06	0.82 ± 0.05	0.93 ± 0.04
test R²	0.995 ± 0.009	0.993 ± 0.001	0.992 ± 0.001

Three different fitting schemes were tested: fSAPT0(1) scales all terms; fSAPT0(2) scales only the two dispersion terms, E_disp⁽²⁰⁾ and E_exch–disp, treated independently; and fSAPT0(3) scales only the sum of the two dispersion terms, E_disp⁽²⁰⁾ and E_exch–disp. The scaling factors were determined over 1000 iterations of multiple linear regression on randomly selected training subsets of the dimer systems, while RMSE and R² were evaluated over the same iterations using test subsets comprising all dimer systems not included in the training subset. Training and test subsets were equal in size.

Not fitted.

Both dispersion terms share a single fitted coefficient.

Table 5. Linear Scaling Factors for SAPT0/aug-cc-pVTZ Energy Termsa.

Table 5. Linear Scaling Factors for SAPT0/aug-cc-pVTZ Energy Terms^a.